When only one thing works

Check for yourself

One of the benefits of travel is seeing with your own eyes. Travel to China, and it's obvious their payments and transport systems are about 10 years ahead of our own. Not that we would read about that locally. China is bad, you see. 

Travel to Britain, and it's obvious (at least to me) that things are not going that well. Not because anything is as terrible as it is portrayed in the press. It's the same. The trains from London to the North of England are the same ones I used to travel on 20 years ago. Literally the same locomotives with the same names. The tube in London is the same (except the shiny new Elizabeth Line). They work fine, but they are old and there has been zero progress.

It’s not a particular surprise that most people don’t see decline either because decline is relative and our politicians can hide behind it. The status quo is failure. Everything should get better, always. Always more energy, always more production, always more leisure time. They are the measures.

It’s also telling that GDP in Britain has risen on average by 2.7% annually over the last decade but they have fallen massively behind. Why? Because GDP itself is a fraud, it's a pretend number that bears no relation to population. Only per capita counts and even then the number is made up (please see the Village Watchman from 2023). Britain has fallen massively behind over the last decade; GDP per capita is still at 2019 levels. 

So it was last week, with the “Deaths from inaction have no face”. We ought to compare against what could have been. We should already be in space, we should already have harnessed the limitless energy of the universe. We should also be using energy as money, since it is the only true measure of progress and the only thing that has equal measure across the universe.

The United Kingdom is many kilojoules poorer than it used to be. On its current trajectory Australia will be too. Getting out of the energy hole in Australia will be a lot easier here than for the UK and we will get out of it, but not yet. It will take the relative growth in wealth of another country to become obvious before we take meaningful action. 

There is an argument that in the Age of AI only energy will matter because everything else will be instantly produceable. It's become like that in software, we are very close now to just imagining software and the AI delivering it. The physical world will follow on a longer timeline; machines will produce anything you want and simply work out how to do it. 

Only energy will matter. Is it a coincidence that Microsoft has effectively sequestered Three Mile Island? Google has its own nuclear deal and Elon Musk has reportedly signed up Rolls Royce to acquire their modular nuclear reactors? 

Government statistics, GDP, inflation, all stories. You can see it in how many Brits believe their country is richer than Australia or the US. They believe the story. 

It is simple. How much energy do you have? Either as a country or as an investor.

Broken Calculator

The broken calculator problem is a maths puzzle where you must reach a target number using a calculator with specific broken keys (numbers or operations). It tests logical thinking and numerical flexibility by requiring participants to find alternative methods, like substituting subtraction in addition to circumvent broken buttons. It’s even in the NSW maths curriculum. 

The broken calculator can get more and more complicated, fewer keys or the removal of specific scientific functions. 

A researcher in Poland, Andrzej Odrzywolek, revealed this week that one single function and the number 1 can generate any standard real function. It is called EML, exponential minus log. His broken calculator would need only two buttons. 

With that one function on a scientific calculator and the number 1, you can generate anything else the calculator could have computed. 

He also helpfully charted the path from EML to other functions in this rather pretty diagram which lays out the number of steps from one function to another. 

Lots of the twitterati were chiming in and claiming that this is all ‘obvious’, but I suspect the best discoveries are obvious only in retrospect. His full paper is here. It’s not peer reviewed and I can’t find anything meaningful in the news about it, so it might be nonsense. These days however you can check if something is nonsense. So, I checked. Those of you looking for projects in Claude Code might do the same.

1. Give it the paper (and the supplementary paper)

2. Then ask it to write code that will produce Pi in terms of EML and 1. 

I would say it was non-trivial to do because Claude Opus took about 10 minutes to work it out and check it, with me prompting it to explain it in the way I wanted, but it seemed convinced.  

π = E( E( E(1, E(E(1, E( E(1, E(E(1, E(E(1,

 E(E(1, E(1, E(E(1,1),1))), 1)), E(1,1)) ), 1)), E( E( E(1, E(E(1, E(1, E(E(1,-1),1))), 1)), E( E(1, E(E(1, E(1, 

E(E(1,-1),1))), 1)), 1 ) ), 1 ) ) ), 1)), E( E(1, E(E(1, E(1, E(E(1,1),1))), 1)), E(1, E(E(1, 1), 1)) ) ), 1 )

Not exactly helpful for a human but it works. For a computer, to only have to work with a single function and 1 might be very helpful. Claude Code was very excited about this ‘discovery’. The paper is off for academic review and he’s either getting a Nobel prize or is being banished to undergraduate lectures forever. 

One of the things about the new era of AI that I like the most is that you can engage with things that would otherwise be way over your head. In this case the maths is over my head and the code required to check it is over my head but I can still use one to check the other and get an explanation of what is going on at a level I do understand. 

Goldman Sachs

The JP Morgan Bitcoin ETF launched this week. A solid start with positive inflows every day and they are heading towards $100 million quickly. Likely the product will be a slow grind as their advisers trickle client money in. 

It was enough to push Goldman Sachs into action. They have filed a Bitcoin Income ETF. The product will hold bitcoin and sell bitcoin options to generate income. It should outperform bitcoin in bear markets and underperform bitcoin otherwise. I’m not keen since it looks like a product designed to generate income elsewhere for the bank. Specifically, Goldman will probably be writing the options themselves, selling them to the fund, earning fees on both sides of the trade. Complexity for complexity’s sake I would say, but there is clearly demand for these ‘clever’ products, so good luck to them. 

Anyway, here are the Goldman slides from 2020 (that leaked) in which they told clients Bitcoin wasn’t an asset class, couldn't generate income and wasn’t investable. 

Welcome aboard. 

Nobody uses it

Not entirely cause for celebration that Iran has chosen to seek payment in bitcoin. Still, it's worth asking the question. Why? 

One of the best things that ever happened to Bitcoin was the seizure by the Biden administration of Russian Central Bank USD in overseas bank accounts. It was wildly stupid because it proved to the rest of the world, and China in particular, that any USD or US Treasuries they hold are worth as much as America decides they are worth. One of those outcomes is zero. 

As a consequence China accelerated their unloading of US Treasuries and began moving into other assets, notably gold. 

Recall too, every movie from 1950 - 2000 had the baddies accepting USD in suitcases. Everyone wants the liquid thing that they can sell and move instantly and will be accepted everywhere. It was once US dollars and it isn’t anymore. 

Some will say this is bad for bitcoin. If we reflect for a moment though; the internet in Iran has been down for months, the infrastructure is slowly being destroyed. They are surrounded by aircraft carriers. They are being bombed into submission. All their communications are being jammed or scrambled. Nothing works. 

Except bitcoin. 

Euro-Trash

It’s been a while since we heard directly from Christine. She was back this week talking up one of her old buddies Mark Carney. He has been nominated as one of the top 100 most influential people in the world by Time Magazine.

“A Rock Star central bank governor”? I wonder. A better way to judge success is what is left behind afterwards. Whatever the circumstances of his time at the Bank of England and the Bank of Canada both of those countries' economies are now in far worse states than when he took those roles. Carney complained that Brexit was the cause of many of the issues Britain faces, but he was hugely influential at the time. He did not make the case for ‘Remain’ strongly enough. He lost, and then Britain lost too. To paint his tenure at the bank as success is just plain wrong. 

It’s telling to me that she describes him as the “George Clooney of finance”. I agree. That is exactly what Carney is. Much of his success climbing the ladder of politics and bureaucracy is down to how attractive and charming he is.

In his book “Fooled by Randomness” Nassim Nicholas Taleb describes a policy of choosing the least attractive surgeon possible when selecting a doctor for an important operation. The reason is that person is likely far more skilled, they have got there on talent alone, not charm and good looks. What matters in very important situations is only skill and competence; one should select purely for that. 

Once you see it, you see it everywhere. A reasonable heuristic might be to be suspicious of those in public office or politics who are very attractive because there is very strong evidence to suggest that’s a large part of how they got to where they were. Gavin Newsom of California springs to mind.  

I believe Mark Carney to be wrong about almost everything. He’s a zealot who does not understand the wealth creation mechanism, he left Britain in a terrible state. He could have done better. He will do the same to Canada. But it doesn't matter because he is attractive and charming and people do like being around him and basking in that charm. Consequently, important people like Christine Lagarde are willing to prostitute their own reputations to cement his friendship. 

I found it particularly thrilling for both of them that the piece was sponsored by Rolex too. Brilliant stuff.  

AI Footnotes

As a matter of interest (because I know many of you are using it), I pass this weekly note through AI to check spelling and facts etc. As the AI gets better, it also gets worse. I pasted all its suggestions below, I took two of them, the hard errors. 

The rest though are mostly equivocations. It does not want to take a position, simply to soften everything, not offend. The comments are from GPT 5.4. GPT 4.5 had far more of a personality and was a much better writer. The latest models are like HR staff, spending hours not offending and ‘crafting the message’ to the point that nobody wants to hear it. 

I think I would have hired GPT 4.5 and I would fire GPT 5.4, it’s weak.